SOURCE: Austin, N. J. E. “Strategy: The Collection and Evaluation of Strategic Intelligence.” In Ammianus on Warfare: An Investigation into Ammianus's Military Knowledge, pp. 22-41. Bruxelles, Belgium: Latomus, 1979.

[In the following excerpt, Austin considers Ammianus's understanding of military intelligence.]

It has been shown above [in N. J. E. Austin, Ammianus on Warfare: An Investigation into Ammianus's Military Knowledge] that Ammianus, as part of his duties as a protector domesticus, was personally involved in the collection of intelligence of a military nature which was of use more in the long term than immediately. He is also familiar with the duties of other individuals instructed to carry out similar tasks.

First then, autobiography. Ammianus' Corduene exploit was pure intelligence work: the background is clear enough. During the early part of 359, the protector and ex-rationarius Antonius defected to the Persians, taking with him a considerable amount of classified information particularly about military matters in the East¹; it took place at a time when Constantius was committed on the Danube frontier and when serious and delicate peace negotiations with the Persians were being conducted². Antonius' information was valuable enough to determine Sapor to invade. Raids had already been taking place (18.6.9) before Ursicinus and Ammianus reached Nisibis (= Nusaybin) and shortly afterwards retired through Amudis to Amida (= Diyarbekir)³; this is the moment when the Roman negotiators at the court of Sapor became aware of preparations for something bigger than mere raiding, and secretly sent a message in a kind of allusive code to Ursicinus at Amida. Their message lent substance to earlier intelligence gathered from speculatores and deserters, who had reported that the Persians were speedily making preparations for war (18.6.8)⁴. Ammianus must have seen this message and taken part in its interpretation, and would therefore have known the background leading up to its despatch. It was very necessary now for fuller information to be acquired in order that proper preparations might be made for the defence of Roman territory, and for this purpose Ammianus was sent to Iovinianus, the Persian satrap of Corduene, who was known personally to him; he was the same age as Ammianus and favourably disposed to Rome, having been brought up in Antioch. A reliable centurion accompanied him⁵. The outward journey was not made along the main communication routes but per auios montes praecipitesque angustias (18.6.21).

Having made contact with the satrap and privately explained his mission, Ammianus was sent on to some distant high mountains⁶, from where after a wait of two days, he was able to see (at a distance of 50 miles!) the immense forces of the Persians crossing the Zab bridge (18.7.1), bypassing Nineveh at a point south of the city, and making for Singara (= Sinjar)⁷. Ammianus assessed their rate of progress at this crossing, taking into consideration the fact that they would take at least three days to do so. This would have indicated an enormous number of troops, though it is impossible to guess at the figure more than to say that Ammianus claims that the Persian forces at Amida a few weeks afterwards numbered 100000⁸. He returned to Iovinianus, took a brief rest and returned to Ursicinus; his report was used to correct earlier reports of the progress of the main body of the Persian army (18.7.2).

Some aspects of Ammianus' account of his actions and findings in Corduene are suspect, as noticed by Thompson⁹ among others: that he was able to see fifty miles and still recognise the faces and dress of the various notables¹⁰; second, that the Persian army contained (visible) contingents from different ethnic groups under their control; and third, that the Persians sacrificed in the middle of the bridge over the Zab. The first of these need be no more than a topos, and thus no more than a convention that would have been expected by his audience. The second could easily have been the result of a conflation of several different intelligence reports, made later at Amida—he did correct earlier versions, and in any case, being on Ursicinus' staff would have had access to other intelligence reports. Conflation is further indicated by Ammianus' knowledge (probably derived from prisoners or deserters at one of the sieges) of the names of the ethnic contingents. The third though is quite wrong, since Persian religion did not permit the sacrifice of living creatures¹¹. These are, however, details and do not affect the fact that Ammianus brought back valuable strategic information for action by Ursicinus' staff¹².

The immediate result of Ammianus' report of his discoveries in southeast Corduene, and of its collation with others coming in, was the prompt despatch of instructions on both the civil and military front to Euphronius as rector of Mesopotamia, and Cassianus as dux, to take appropriate action: the movement of the country population to safer areas and the abandonment of Carrhae (= Harran) because of its weak walls (18.7.3). Ammianus states that these orders were quickly carried out and a strategic defence policy implemented—that of destroying all the ripening crops along the direct route west, and most of the fodder needed for pack animals¹³. The diplomatic appointment Sabinianus, based now at Edessa, does not seem to have taken a very active part in all this activity; we may assume that Ursicinus was in the main responsible for the planning of the defensive moves. Working now on all the information that must have been coming in all the time, Ursicinus deduced that Sapor was aiming to push westwards, cross the Euphrates, and move in on the Syrian provinces¹⁴. To supplement the scorched-earth policy mentioned above, he organised parties of tribuni and protectores to set up suitable defences along the west bank of the Euphrates (18.7.6). The scale of the defensive measures employed by Ursicinus and his staff implies that their understanding of Persian strategic aims was on a high level and shows that the fullest possible use was made of the intelligence material that had been obtained. The use of this material and the effectiveness of the Roman response (in the initial stages at least) indicate that the links between intelligence, planning and operations were working as they should. Ammianus as a direct participant on Ursicinus' staff, and possibly given more information on the operational side by Cassianus, was able to follow the moves and counter-moves accurately and to transmit them in a comprehensible form. There are no real difficulties about his account until the point at which the Persians arrived at Bebasa (= Tell Chagar Bazar) and delayed there for a time before unexpectedly turning north (18.7.9)¹⁵.

Up to this point then, the Persian general plan was to move more or less directly westwards with the intention, one imagines, of crossing the Euphrates at Carchemish-Europus or perhaps a little further south near Hierapolis. The implementation of the plan was made more difficult by the Roman removal of the inhabitants of places around Carrhae to more secure areas, the burnt-earth technique and the defensive measures taken on the banks of the Euphrates. The final abandonment of the Persian plan was brought about by the discovery of unseasonable flooding of the area through which the Persian army was to pass¹⁶. The consultations held by Sapor with the defector Antoninus at Bebasa resulted in the adoption of an alternative plan, which Ammianus apparently does not understand quite so clearly. This of course may be due partly to the confusion created by his own adventures. He does however supply sufficient material for the real Persian plan to emerge without undue difficulty. The real alternative seems to have been as follows: from Bebasa, the Persians would advance north to Amida, and further north to Ziata (= Ammaneh), about 50 km. away, which had been captured during the course of the siege of Amida (19.6.1); they would then turn north-west towards Mazara (= Mezre), past the lake Hazar Gölü (about 100 km.), and then either turn roughly south-west to the Euphrates crossing at Ad Aras (c. 60 km.) where Melitene (= Malatya) and thus Syria is a comparatively short march away; or else go further north and cross the Arsanias (= Murat river), to avoid the floodwaters below the confluence of the Euphrates and the Murat, and thus more easily get across the Euphrates¹⁷. This route would take the Persians well away from the flatter area flooded by the river, and higher up its course, where crossing would be much less difficult for a large army. Water and food supplies would be no more of a problem this way than they would anywhere else. This scheme of operations was of course delayed significantly by the lengthy siege of Amida.

Ammianus (and probably therefore Ursicinus) however did not see the significance of the capture of Ziata in the overall picture of the campaign. If he had grasped this point with any clarity, he would not have been misled into thinking (or believing) that Antoninus really advised Sapor to move northwest from Bebasa to the castra duo praesidiaria Barzalo and Claudias (18.7.10). It is possible that the certae speculationes of 18.8.1, who brought Ursicinus the information regarding the Persian change of plan, were only able to indicate a northward turn, the logical extension of which meant that the Euphrates would now be crossed somewhere near these two points. It is also just possible that this information was deliberately relayed to the Romans as part of Sapor's well-laid plans for the area, specifically to disrupt Roman contingency planning. The intelligence brought in was translated into action by Ursicinus and his staff who were on the point of setting out for Samosata (= Samsat) to break down the bridges to the north of the previously estimated crossing point, i.e. at Zeugma-Apamea and Capersana (18.8.1). Their attempt was frustrated by the arrival at Amida of almost 20,000 fast-moving Persian troops under Tamsapor and Nohodares, and Ammianus and Ursicinus were separated, the historian being forced to take shelter in Amida. Ammianus' own adventures which follow the siege have little to do with the strategic intelligence background of the campaign except perhaps to underline the element of incomprehension noted earlier. After his escape and arrival on the banks of the Euphrates (19.8.9), perhaps near Barzalo or Claudias¹⁸, he and his two companions were terrified by witnessing the pursuit of a number of Roman cavalry by a large Persian body¹⁹: this indicates the presence of advance Persian units already along the line of the Euphrates, presumably clearing the way for the main forces. They do not, however, appear to have reached the area around Melitene, where Ammianus and his companions arrived after a journey along routes which were not likely to bring them into contact with any Persians²⁰.

The preceding analysis has dealt with Ammianus' treatment of the collection, handling and use of strategic intelligence material from an autobiographical section of his work. But a better picture of his understanding of the principles involved will emerge from the discussion below of several non-autobiographical incidents.

The first concerns Julian in early 359, and forms part of the account of his general reconstruction of Gaul and the restoration of territorial integrity. While he was wintering in Paris, he was working on plans to subdue those pagi of the Alamanni which had continued to exhibit hostility after the battle of Strasbourg some eighteen months previously (18.2.1), and which he had been handling during the less important campaigning of 358²¹. He was however unable to decide what type of campaign this should be, due to a lack of satisfactory relevant information²² though he intended in any case to operate that season on the N. Rhine frontier²³. To acquire this information, he despatched secretly one Hariobaudes, a tribunus uacans, legationis specie to the king Hortarius, but in reality instructed to go further into hostile Alaman territory with the general aim scitari … quid molirentur (18.2.2). His personal qualifications for undertaking this mission are particularly noted by Ammianus: outstanding reliability and intrepidity (both of which would be required, as on the accuracy of his report depended much of the conduct of Julian's campaign; without successful penetration of the enemy's preparations it might not be sufficiently reliable), together with a complete fluency in the local dialect²⁴. Hariobaudes returned, perhaps to Vingo, and the immediate effect of his report was that all Julian's forces were quickly assembled at Mogontiacum, but did not throw a bridge over the river immediately, perhaps because the report indicated that Alaman troops were close by in deterring numbers (18.2.9).

Not having been an onlooker or participant, Ammianus does not know the precise content of Hariobaudes' report, but in large measure it can be inferred from the events that follow. Thus it appears that Suomarius and Hortarius were being subjected to pressure to refuse the Romans access through their territory and to allow the free passage of forces hostile to the Romans through their territories (18.2.8 and 13). The report therefore indicated that a marked level of preparedness had been attained by the Alamanni and that diplomatic activity was in progress: the fact that a feast was to be held at Hortarius' encampment may well have also been known in advance through Hariobaudes' visit to that king first, otherwise there seems to be no reason for the sudden assault on Hortarius' encampment very soon after, in spite of Ammianus' specific casu and ex improviso (18.2.13). If one accepts that the Romans knew about this feast beforehand, then it is again easily possible that Hariobaudes took a suggestion from Julian to Hortarius that all those reges, regales and reguli be invited to a feast, thus effectively detaching them for a time from their main sources of military support. The swift move of the Romans down the Rhine²⁵ and the subsequent night attack were therefore designed to create an atmosphere of confusion and panic among the Alaman leaders, without being particularly successful in bringing about their deaths, which was probably the intention in planning the raid. The purpose of all this was to keep surprise on the Roman side. Ammianus emphasises this state of panic and the efforts of the Romans to maintain it as long as possible until a full-scale invasion was already under way²⁶.

In this campaign it is the position of the king Hortarius which perhaps provides the key to the action. It was through his territory that the tribune Hariobaudes travelled, it was at his encampment that the night raid took place, it was through his regna intacta that the Romans marched to get at hostile territory. The whole strategy of the campaign was therefore largely dictated by his attitude, and by the information that he perhaps gave Hariobaudes to supplement what the latter had obtained himself. Without the assurance of tacit connivance or explicit assistance from Hortarius—one of the obvious reasons for the intelligence mission—Julian's long-term planning for the year's fighting would have been seriously handicapped. Ammianus then does not in fact appreciate or indicate the full purpose of Hariobaudes' contacts with the Alaman king—the importance of the contacts can be sufficiently demonstrated by the fact that no other action was contemplated or undertaken by Julian during 359. The account of the campaign is summary and in many ways unsatisfactory; it is safe only to make the suggestions outlined above. Ammianus' rather poor sources of information did not provide him with any dates (thus we are unable to infer whether this was a blitzkrieg, or most of a season's campaigning); nor is there anything in the way of place-names after the departure from Mogontiacum (18.2.9); that it occupied some time is suggested by the mention of Roman penetration and devastation as far as Capillacii-Palas, on the frontier between Alaman and Burgundian-controlled territory (18.2.15)²⁷, which in turn suggests that the grass-roots of anti-Roman subversion were to be attacked, though total destruction was perhaps impracticable.

The mission of Hariobaudes can be called successful from the fact that the information he returned with, and the complementary information that must have come from Hortarius, were found to be accurate and reliable, and of direct relevance to the planning of the campaign, as its outcome shows (18.2.16-19). The effect of the various intelligence factors is however clearly visible in the overall success of the campaign as a whole, particularly on the strategic level. Ammianus does not seem to grasp all the implications, but is still able to avoid making Julian's movements appear purposeless, in spite of defects in his basic information.

A second incident, contemporary with the previous one, concerns Constantius and the Limigantes. Early in 359, while still wintering at Sirmium (= Sremska Mitrovica), Constantius received information about renewed restlessness amongst this people, who were an offshoot of the Sarmatae; as Roman dependents, they had been allocated only the year before territory across the Danube in what is now southern Hungary. In the view of Constantius it was essential to quell any disturbances as quickly as possible, and also to prevent the Limigantes from crossing the Danube while it was still low before the spring snow melted and raised it to flood level; he set out at the beginning of April²⁸ and on arrival at Acimincum (= Slony Kamen) in Valeria, encamped on the banks of the Danube, at a point where the actions of the Limigantes could be observed. Two tribuni each with an interpreter were sent across the river with the aim of finding out the reason for the contravention of the conditions laid down the year before. This mission was far more open than Hariobaudes' since there was no secrecy about their despatch and no disguise of their identity, as is indicated by their going with interpreters. The knowledge of the language that the latter would have would tend to imply a degree of familiarity with the conditions and circumstances of the Limigantes, and this would materially assist the tribunes' own assessment of the facts. It is likely then that the party was to report on the morale and aggressiveness of the Limigantes as well as their military preparedness, otherwise the mission would have had little real purpose. The first of these seems to have dominated their analysis of the facts as presented to Constantius: they had been taken in by the evidently plausible presentation of the reasons for the restlessness and promises for the future²⁹, and that influenced Constantius' judgment (as it was intended to), and so the Limigantes were granted entry into the Empire.

The whole mission is open to criticism for its lack of success. In his relief and gratification at the apparent mildness of the reply given by the Limigantes, Constantius perhaps slackened proper security precautions which should have been maintained in view of the restlessness which prompted the expedition originally. In allowing the tribe to cross the river, and then preparing to address them while protected only by a small bodyguard on the spot and light-armed legionaries on ships patrolling the river³⁰, Constantius exposed himself to considerable danger and only escaped in the fighting that followed by his promptness in leaping into his horse and galloping away (19.11.11).

There had therefore been some misconception of the intentions of the Limigantes by the intelligence mission, an error which gravely endangered the life of the emperor himself in its consequences. Ammianus, however, puts the blame largely on the cohors adulatorum who apparently played on Constantius' awareness of the shortages of effective military manpower in the empire: but it seems clear that their responsibility was only contributive³¹. The principal failure lay in the wrong interpretation of the facts: the mission would have been able to see that in the context of the restlessness, which had so disturbed Constantius that an expedition under his personal command was prepared, any reasons given by the Limigantes for it should have been viewed with marked scepticism; and there is a further factor in that the Limigantes were armed missilibus … gladiis et uerrutis³² when the attack was made on Constantius: if the Limigantes had been preparing to cross the river, the arms would have been on their persons, and thus visible to the Roman mission, who should have reported to Constantius with the obvious emphasis³³. This point is of tactical rather than strategic importance, but it should have shown that for the long term this was an indication of future action. The apparent submissiveness seems to have affected the nature of the mission's report: the fact that the Limigantes were armed was too little stressed for it to have had its proper influence on Constantius' handling of the situation, or else it was under-emphasised during the process of evaluation and discussion. The very openness of the mission may well have been another reason contributing to its ineffectiveness: the Limigantes gave a reply that was likely to ring acceptably in Roman ears. There was in fact no strategic outcome to the strategic information required by Constantius: it was all over in a single tactical exercise which drastically simplified the matter by destroying the cause. Ammianus' account of the mission of the tribuni and their interpretes suggests that he was well aware of the shortcomings and distortion given in the interpretation of their report: the wide-ranging criticism in 19.11.7 containing allusions to economic conditions and recruitment difficulties on an empire-wide basis, indicates among other things his own dissatisfaction with the conduct of the mission and the subsequent use made of its information. His presentation of these facts goes some way towards demonstrating his knowledge of the connection between intelligence activity and the use made of such information in the planning of operations. Some evidence on the kind of source that Ammianus derived his information from can now be seen. He shows less interest in the actual report of the mission than in its interpretation: interpretation and evaluation of intelligence must take place in a headquarters, and thus Ammianus' rather stronger emphasis on the interpretation suggests that some at least of his information for this incident may have come from a person who was present at the discussion concerning the report, or from records of such a discussion. In this connection, the agrimensor Innocentius (19.11.8) could have been his informant³⁴: the otherwise inexplicable notice given to him as the originator of the plan of having a naval patrol on the river during Constantius' address to the Limigantes points towards his having been a source for these events—it is conceivable that such a man, if he did suggest a plan of this nature, would have been present at some relatively high-level meetings where the course of action would be mapped out.

There are some similar missions which are worthy of mention. During the course of 366, the beginnings of the long and disastrous contacts with the Goths took place, following their support for the usurper Procopius (365-6)³⁵. Valens had already been made aware of the conspiracies and preparations for war being made among the Gothic peoples on the lower Danube, by means of intelligence reports provided by the various duces in the area³⁶. These reports were presumably included in routine security reports which would be examined by the emperor's senior military staff as a matter of course. Ammianus' mention of their existence here seems to be a firm pointer towards his knowledge of the methods of the collection and handling of strategic intelligence; this is strengthened again by his statement that the reports of several duces contained such information, and from their collation and evaluation a clear idea had been gained of where the Goths were likely to try to effect a crossing. These points suggest that Ammianus was closely acquainted with the organisational steps required for the handling of intelligence material, long before the real planning of a campaign began³⁷. The revolt of Procopius burst into the open on 28 September 365, and it provided an excellent opportunity for the Goths to exploit: acting on the information he had already received, Valens had sent a number of troops to strengthen the garrisons in Thrace, via Constantinople³⁸—these Procopius suborned, thus effectively weakening the Roman position on the Danube; he also called on the Goths for help, emphasising his relationship to the family of Constantine and the break with the old dynasty that the accession of the Pannonians Valentinian and Valens represented. In this way a number of Goths were ‘legally’ admitted into the empire, and presumably utilised by Procopius and his ‘successor’ Marcellus, until the revolt was finally suppressed at the end of May 366.

It would seem that Valens regarded their participation in Procopius' attempt as an act of war, and determined to engage in a punitive expedition. In the event this occupied three years and still ensured the submission of only one section of the Goths, that under Athanaricus³⁹. Valens' magister equitum Victor was therefore despatched to the Goths, once Procopius had been removed, in order to find out the precise reason for their action. Undoubtedly, in view of Valens' immediate organisation of the expedition against them, he was to make the most of this opportunity to assess the size of the threat that the Goths presented (27.5.1). The reply of the Goths, incorporating as it did a reference to the relationship of Procopius to the family of Constantine, and Valens' reaction to it⁴⁰, indicate the likelihood that Victor's report must have contained sufficiently serious facts and figures for the campaign to be mounted without delay. Ammianus was now no longer in the army, and was not able to get at the exact terms of Victor's instructions; he was not really in a position to provide any more information than he already gives; one can only deduce the nature of Victor's report from the account of immediately subsequent events.

By the end of 369 the Goths were ready to treat, not so much as a result of military defeat, but more likely from being deprived of their usual sources of food⁴¹. Their frequent legationes sent to Valens resulted in the despatch of Victor, this time accompanied by the magister peditum Arinthaeus (27.5.9), to examine their economic conditions. The reliability of their assessment is emphasised by Ammianus here and their report must have provided the information that the Goths would be unable to present any great threat on a military level for some years.

It is noteworthy that very senior military men were employed in these missions into Gothic territory. Valens clearly regarded the collection of reliable strategic intelligence as vital for his plans to secure the frontier on the lower Danube as a whole, and for this campaign in particular, especially as he spent three years conducting it⁴². In this aim he was fairly successful, as the frontier was not breached in the area until the period before Adrianople, eight years later. Victor had been able to see that the Goths were arming, and in spite of their claims of having supported the man they considered the legitimate successor to Julian (27.5.2), were obviously preparing to invade Roman territory. Valens' reaction to the report was to invade promptly the area that used to be the old province of Dacia. Bad security (the Goths knew of the invasion before it occurred, 27.5.2) and a lack of proper strategic planning on the Roman side led to a Gothic withdrawal into the montes Serrorum (= Transylvanian Alps): inaccessible to the Romans. Ammianus misunderstands the nature of the Gothic moves in ascribing them to fear of the splendid legionary equipment (27.5.3): what must in fact have happened was consonant with Roman warfare against the peoples of Central Europe then—no direct confrontation, the enemy remaining elusive and not combining into anything larger than bands. This was merely a defensive form of the methods normally adopted for offensive raiding⁴³ and forced Valens into a three-year action which the Gothic capacity for open warfare in this period did not entirely justify. The campaigning season of 367 was fruitless, and thus Roman strategy had to be modified by limitation to destruction of food supplies for the Goths (27.5.4), an aim perhaps furthered by the unusual weather conditions of 368 (27.5.5). Operations further east in 369, perhaps in concert with more in old Dacia not mentioned by Ammianus, completed the reduction to starvation level of the Goths, carefully confirmed by the second visit of Victor, with Arinthaeus.

The intelligence report of Victor then brought about strategic action on two aspects, limited though the planning may have been at the beginning: the first suggested a rapid, direct assault on the Goths to break up their war organisation and conspiracies; on their retiring to the mountains, the second suggested the slower method of starvation⁴⁴. But it is clear that general action was needed to maintain reasonable security on the lower Danube frontier. The form that this action took depended on the nature of the information brought back from Victor's mission.

One further example of the collection of strategic intelligence and its subsequent handling can perhaps be found in Theodosius' North African campaigns during the years 373-5 against the rebel Firmus (29.5). Shortly after an unheralded arrival at Igilgilis (= Djidjelli), Theodosius met the comes Africae Romanus, who was sent off to organise a garrison network and frontier defences in Mauretania Caesariensis. Once Romanus had left, Theodosius issued orders for the arrest of the tribunus Vincentius, curans Romani uicem and sent Gildo, a brother of Firmus' to carry it out. Vincentius seems to have got wind of his impending arrest and ‘disappeared’. Theodosius advanced in a generally westwards direction, recapturing a number of towns which had been destroyed, until he reached Caesarea (= Cherchel) where he stayed a while to consolidate his position and plan the next phase of the operations. Encouraged by the official propaganda that was being disseminated (29.5.19), Vincentius emerged from hiding, making use of the temporary peace to get away from an area held by Firmus⁴⁵; accompanied by some members of the provincial civil administration, he made his way to Theodosius' headquarters. It is possible that when he had heard he was about to be arrested, he had defected to Firmus as had a number of other Romans and some army units⁴⁶. If in fact he had defected, it would explain why he was so conversant with Firmus' duplicity in secretly planning and preparing to throw the Roman army into complete confusion. The intelligence information that Vincentius and the other officials brought in was of such value that they were gratanter uisi atque suscepti (29.5.19), which would hardly have been the case otherwise, considering Theodosius' usual treatment of deserters and waverers⁴⁷.

Theodosius' campaign was saved from becoming a long-term defensive action by the strategic information brought in by Vincentius. It was only at this stage of the campaign that he was made fully aware of the extent of the support, overt and covert, that Firmus was able to command. This may not have been altogether clear before Vincentius returned to the Roman side with high-grade information about precisely this support and how it was to be utilised by the rebels to create chaotic conditions for the Roman army to operate in⁴⁸. It is noteworthy that it is only after Vincentius' return that Theodosius' planning and action became directed towards clearly-defined objectives. These do not emerge from the earlier part of Ammianus' narrative. Theodosius' immediate departure inland from Caesarea to Sugabar (= Zaccar) and Tigaviae (= Kerba, on the river Chélif), indicates that he realised that unless full military discipline was restored and the areas ‘held’ by demoralised or dissatisfied troops were brought under control, his chances of putting down the revolt were slim⁴⁹. The policy now adopted perhaps accounts for the reference by Ammianus to the destruction of fortified farms: it being particularly necessary to destroy every form of foothold and shelter that the rebels might be able to use⁵⁰.

The result of Vincentius' information was that Theodosius could now make more definite progress towards re-establishing Roman authority. His strategy, a standard one in all forms of guerrilla warfare, was to weaken on all possible levels the areas of security and supply of his enemy. This took time, as Firmus was not disposed of until some time in 375: the explanation seems to lie in the limited number of troops at Theodosius' disposal as a rapidly-moving strike force (29.5.29), as opposed to garrison-duty troops, and in the widespread indifference to Roman success or failure engendered by years of exploitation by Romanus or his associates (27.9.1-4; 29.5.50). Theodosius acted on three levels—military, propagandist and administrative: major successes in the first two of these are represented by Firmus' sudden abandonment of the army so carefully built up (29.5.33), and the defection of Igmazen, king of the Isaflenses (29.5.51 ff): in the third by the establishment of reliable praefecti over the tribes brought back to the Roman side (29.5.35). All of these are very clearly understood and described by Ammianus during the campaign narrative, as their results can be so easily seen as the logical outcome of the intelligence received and the planning based on that intelligence. The Vincentius episode is one in which the intelligence (of both strategic and tactical value) was unsolicited and came in purely by chance, and forms a contrast with the incidents discussed earlier, where most of the information was purposely sought out. One could go so far as to suggest that Theodosius' contacts with a number of the Mauretanian tribes during the course of the campaign (29.5.33), apart from the reasons advanced by Ammianus (pacification, and retention or restoration of allegiance), were partly also intelligence missions. Theodosius would undoubtedly have expected the men making the contact with the tribes to have briefed him on their morale, rumours, local feeling about Firmus, and so on, and from this information been able to draw up a reasonable plan for internal security.

In the same North African context, an example occurs of a significant failure either to collect or to interpret correctly intelligence items of major strategic importance. This is to be found in the ‘prelude’ to the revolt of Firmus, at the point where Ammianus deals with the immediate causes of the revolt (29.5.2): the very serious mishandling of the succession to Nubel. The chief military official in the area, Romanus, appears to have failed to grasp the consequences of this mishandling and of the state of military preparedness of the Mauretanian tribes and their supporters. This latter point suggests that dissatisfaction had been rife among them for some considerable time⁵¹. For these errors of judgment a large measure of blame must attach to Romanus' intelligence sources, if not on the comes himself for so obviously failing to ensure their efficiency. In addition, there had been reports and correspondence between Romanus and the imperial court, and letters from Firmus himself (29.5.2); all of this must have been going on over a period of several months at least, during which it is reasonable to assume from Ammianus' account (29.5.3) that Firmus was preparing for the break with Rome. Since a fair number of the local tribes joined him⁵², it seems surprising that information concerning the preliminary rumblings of discontent and the preparations for a rising did not reach Romanus, or if it did, that it was not acted upon sooner by the local commander. Given the number of tribes involved, it would appear that the latter is the more probable. Once the outbreak had occurred, Valentinian himself acted. While Ammianus' account does not give this interpretation in quite so many words, it does however supply sufficient information for much of it to be deduced: he is quite clear, though, about the general negligence and corruption of Romanus' interested maladministration, and this interpretation of an aspect of Romanus' incompetence must therefore remain valid. Though Ammianus was generally aware of the fact that strategic information is collected over periods of varying length before the mounting of expeditions, in this case he missed (or was not informed about) the value to the Roman authorities of this kind of strategic intelligence. It is of course possible (as here) that where strategic security intelligence was not directly connected with the overall conduct of a campaign, Ammianus did not appreciate its relevance. In the other cases discussed above, where mention is made of its collection and handling, there was an immediate effect on the planning and execution of campaigns. Here there was not (since failure in this regard was a primary cause of the wide-ranging escalation of the revolt), and may be due to his own military background: in his experience, information that is collected is acted upon reasonably quickly and can be seen to be acted upon.

As a final example of the way intelligence was collected and handled, note should be taken of the areani, an obscure body of troops mentioned in connection with Theodosius' campaigns in Britain (28.3.8). From the brief description Ammianus gives of this organisation, it appears that its members were used for the collection and processing of strategic and tactical intelligence gained during forward penetration in the areas ahead of the frontier system based on Hadrian's Wall, per longa spatia discurrentes. The whole corps was disbanded by Theodosius because of their unreliability, occasioned by their proven venality during the barbarica conspiratio which apparently preceded the incursions of 367-369. But their duties are unfortunately not accorded any detailed description (since Ammianus had already described them in an earlier book), and it is not possible to speculate further on their function. Two modern views, however, permit a degree of insight into what Ammianus says. Stevens⁵³ connects the areani with the area-like appearance of the milecastles along the Wall, seeing area in the sense of enclosed garden patches or sheepsteads⁵⁴ which the milecastles could easily resemble. He then advances the theory that the term areani could have been applied to the occupants of the little milecastles by the same kind of military witty slang that gave the names lilium or cippus to certain types of unpleasant defensive fortifications. Richmond⁵⁵ takes this a stage further: he accepts that areani refers to the appearance of their home bases, but suggests that these bases were also perhaps forts ahead of the Wall, such as Risingham and Bewcastle, and thus areani is a word used in place of the former exploratores on this frontier. His opinion that they were occupied in open frontier patrolling seems less securely based, since too much open patrolling might have prevented the acquisition of sufficiently reliable information about the uicinarum gentium strepitus. The failure of the areani to report such strepitus was partly the reason for their being disbanded by Theodosius.

Notes

1. 18.5.1, … et utriusque linguae litteras sciens, circa ratiocinia uersabatur, qui uel quarum uirium milites ubi agant, uel procinctus tempore quo sint uenturi describens, itidem armorum et commeatuum copiae, aliaque usui bello futura, an abunde suppetant indefessa scitatione percontans.

2. Two missions had been sent in 358 (17.5.15; 14.1). It would appear that high-ranking missions were sent, with the genuine intention of reaching an agreement as any threat in the East would seriously stretch the empire's defence system. The appointment of Sabinianus as a diplomatic and not military commander in the East in place of Ursicinus is a further indication of Constantius' policy for the area (18.5.5). See Thompson, Historical Work, 50.

3. Thompson, Historical Work, 51, criticises Ursicinus for fumbling movements after his arrival in the East, in that he made at once for Nisibis and then fell back on Amida and later on Samosata. There is perhaps an element of hindsight in this, as when Ursicinus reached Nisibis there were only raids taking place, without any other apparent aim than to cause as much confusion for the Romans as possible. The main invasion force (which Ammianus saw crossing the Zab and the Tigris) came only after the thrust to Amida, and nearly caught Ursicinus and Ammianus there. He was not to know at that stage that the Persians were intending to strike at Syria.

4. Ursicinus' position in this campaign of 359 is anomalous, in that he had at the end of 358-early 359 been summoned by Constantius to court, to take up the position of magister peditum praesentalis, as successor to the unpopular Barbatio, and was also to be advanced to the rank of comes (18.5.5; 6.1). The news of impending invasion and the failure of the negotiations caused Constantius to send Ursicinus back immediately to the East before he even reached the court, but without giving him the overall responsibility for the defence of the East. Accordingly he was now, temporarily, subordinate to the sluggard Sabinianus, his own successor as magister peditum per Orientem. See Thompson, Historical Work, 50; Ensslin, ‘Zum Heermeisteramt des spätrömischen Reiches’, Klio 6, 1930, 110ff; Nischer, ‘Das römische Heer und seine Generale nach A.M.’, Hermes 63, 1928, 431.

5. 18.6.20-21. The centurio should probably be a centenarius; Vegetius 2.8; Müller, ‘Militaria aus Ammianus Marcellinus’, Philologus 64, 1905, 593-4: Grosse, Römische Militärgeschichte, 117.

6. These praecelsae rupes could be the mountains stretching east of the Tigris and south of the E. Khabour rivers: their southern slopes overlook Nineveh (= Moussoul) about 50 km. away, and further away, the crossing over the Zab.

7. For Sapor's march, see fig. 37 in Dillemann, HMPA, 291.

8. 19.6.11, where there may be an element of exaggeration for the sake of effect.

9. Historical Work, 7.

10. Dillemann, ‘AMPET’, quotes Cicero, Acad. 2.25; Pliny, NH [Northern History] 7.21; and Solinus, Coll. 1.99 as supporting evidence. On this passage, see also Vogt's sound remarks in ‘Amm. Marc. als erzählender Geschichtsschreiber der Spätzeit’, Abh. der Akad. der Wissenschaften und der Literatur 8, Mainz 1963, 823.

11. Historical Work, 8.

12. For a defence of Ammianus' version of his exploit in Corduene, see my paper ‘In support of Ammianus' veracity’, Historia 22, 1973, 331f.

13. Rosen, 28 and 45, and Dillemann, HMPA, 291, call the burnt-earth policy a form of tactical defence: they thus ignore its long-term effects on the conduct of the campaign which achieved their purpose in forcing the Persians to change their objectives.

14. 18.6.3, in the report of the advice given to the Persians by Antoninus. Ammianus may well have discovered what Antoninus told the Persians from Iovinianus when he was there; see Rosen, 18, and Pighi, Nuovi Studi Ammianei, 151ff. Other background may have been obtained from the dux Cassianus, whose apparitor and rationarius the defector had been; see Dillemann, ‘AMPET’, 103.

15. Dillemann, HMPA, 290ff, identifies this as the site of Bebasa.

16. 18.7.9. For an explanation of this flooding, see my paper in Historia 22, 1973, 332f.

17. This follows the Route no. 1 given in Fig. 17 in Dillemann, HMPA, and also the geographical account given there on 155.

18. Dillemann, HMPA, 155.

19. Ammianus can however appreciate their speed of movement, 19.8.10-11.

20. 19.8.12, per dumeta et siluas montes petimus celsiores.

21. 17.8—the campaign against the Frankish Salii and Chamavi; and 17.10—the attack on Suomarius' and Hortarius' territory, opposite Mainz and Speyer respectively (Norden, Altgermanien, 6).

22. 18.2.1, haerebat anxius qua ui qua celeritate, cum primum ratio tribuisset, rumore praecurso terras eorum inuaderet repentinus.

23. Before the return of Hariobaudes from his mission Julian had already seized and restored Castra Herculis, Quadriburgium (= Qualburg), Tricensima (nr. Xanten), Novaesium (= Neuss), Bonna (= Bonn), Antennacum (= Andernach), and Vingo (= Bingen), together with a number of supply depots. In the account of Ammianus the first four of these are all north of Cologne, the last three to the south—it is clear from their arrangement in the text that Julian was intending to move towards Mogontiacum (= Mainz) and a possible bridgehead there (18.2.4).

24. 18.2.2 fin. Jones, LRE, 622, suggests that Ammianus' remark is scarcely intelligible, unless we assume that most of the many German officers in Julian's army had forgotten, or were rusty in, their mother tongue. This seems most unlikely, in view of the fact that bilingualism must have been very much a standard feature of barbari in the service of the Roman army. This is strongly implied at 14.10.7-8, 16.12.2, 29.4.7 and 31.10.3, all quoted by Jones in a slightly different context, but which militate against his claim of loss of fluency. My contention here is that the import of sermonis barbarici perquam gnarus is ‘fully familiar with the local Alaman dialect’, which would of course be an essential qualification in such an espionage mission, as ‘local colour’ and lack of distinctiveness would prevent any undue suspicion being aroused by his presence.

25. The 40 lusoriae naues used by Julian for this purpose, 18.2.12, were probably a flotilla detached from the Classis Germanica stationed at Speyer. See Grosse, Römische Militärgeschichte; 73.

26. 18.2.14: perculsi reges, metu exhorrescentes, sollicitarum gentium opinione praeuenta, etc.

27. This place may perhaps have been as far from the Rhine/Main confluence as the Bayreuth area, about 250 km. as the crow flies; but it was probably somewhat closer and further south, especially as the later negotiations, 18.12.16, involved Vadomarius, whose domicilium was contra Rauracos, i.e. north of Augst. See Thompson, Early Germans, 150, for discussion about the areas they inhabited. Laur-Belart, in Congress of Roman Frontier Studies 1949, 63ff, suggests that the names Capillacii-Palas refer to a place on the old limes (presumably Oehringen and its environs), following the tradition discredited by Norden, Altgermanien, 85ff. The latter's discussion of these two names suggests a regio further east than the limes but he does not commit himself to a definite conclusion.

28. 19.11.4. For the date, see Seeck, Regesten, 206.

29. 19.11.6. Ammianus, with ‘hindsight’, realises this in the light of what subsequently happened: qui uana quaedam causantes et irrita … See also 19.11.9.

30. 19.11.8. The ships were probably part of the classis prima Fl. Aug. normally based at Sirmium, or the classis secunda Fl., at Graium (= Raca), or the classis Histrica, part of which was based at Mursa (= Osijek), cf. Not. Dig. Occ. 32.50, 51, 52. On several occasions Ammianus makes mention of the river patrol boats, as distinct from boats for pontoon bridges: 17.1.4; 2.3; 18.2.12; 30.3.4; etc. See Grosse, Römische Militärgeschichte, 73ff.

31. Ammianus' view of the court circle is of course coloured by the experiences of Ursicinus, who fell foul of it about this period, and thus is not to be regarded as entirely accurate. For references to Ammianus' view of the court, see Thompson, Historical Work, 54, n. 3.

32. 19.11.11. These are typically German weapons of the period. See Thompson, Early Germans, 112f and n1, and cf. Caesar, BG 5.44.7.

33. In a later crossing (376) sanctioned by the Romans, the Visigoths were disarmed. See Thompson, Early Germans, 118f and n1.

34. See PLRE, s.v. Innocentius (4) for a possible identification of this officer.

35. 27.5.1. Apart from a passing mention of the Goths at 22.7.8 and 26.4.5, Ammianus does not pay this people much attention until their involvement with Procopius. They had sent some 3000 men (26.10.4); Zosimus rather unreliably claims the figure was 10000 Goths, with other barbarian peoples sending further help (4.7.2).

36. 26.6.11, … Valens, iamque fines Bithynorum ingressus docetur relationibus ducum, gentem Gothorum, ea tempestate intactam, ideoque saeuissimam, conspirantem in unum, ad peruadenda parari collimitia Thraciarum. This appears to have taken place after the end of July 365; see Seeck, Regesten, 227.

37. A similar example of security intelligence being handled by local military leaders and subsequently passed on for assessment occurs at 26.5.10, the first reports of Procopius' usurpation to reach Valentinian.

38. 26.6.11. The troops in question were the Divitenses and Tungricani Iuniores (12).

39. In 369, 27.5.9. Cf. ILS 770 for the official agreement.

40. 27.5.2, parui ducens excusationem uanissimam. A discussion of the basis of Procopius' claims can be found in my ‘A usurper's claim to legitimacy: Procopius in ad 365/6’, Rivista Storica dell' Antichità 2, 1973, 187ff.

41. Some discussion of Roman trading restrictions and their effects on peoples over the frontiers can be found in Thompson, ‘Constantine, Constantius II and the lower Danube Frontier’, Hermes 84, 1956, 372ff, and in the same author's Early Germans, 90f, and Visigoths in the time of Ulfila, 19. Cf. Tacitus, Germ. 41, Hist. 4.64.

42. Victor's cool reliability is a feature contributing to the success of these two missions. Ammianus mentions that he was a Sarmatian by birth, but cunctator et cautus (31.12.6) and reliable (cf. 27.5.9; 31.7.1). See PLRE, s.v. Victor (4). Arinthaeus who was probably a Goth (see Waas, Germanen im römischen Dienst, 87) had previously been used in positions requiring some degree of initiative: e.g. in the delicate negotiations with the Persians in 363 after Julian's death (25.7.7), and his spectacular conduct resulting in the arrest of Hyperechius at Dadastana in 365 (26.8.5). See PLRE, s.v. for his career.

43. For offensive German strategy and its connection with food supplies, see Thompson, Early Germans, 145. Zosimus (4.11.2), emphasises that guerrilla warfare took place without a proper confrontation, throughout this campaign. See also Thompson, Visigoths in the time of Ulfila, 6, 19.

44. Non-military action in this context was provided by means of the trade ban Valens imposed, 27.5.7.

45. Warmington, ‘The career of Romanus, Comes Africae’, ByzZ [Byzantinische Zeitschrift] 49, 1969, 60.

46. Cf. 29.5.20; also 21, where Gildo and Maximus, sent out at 29.5.6 to bring Vincentius in, brought a chieftain of the Mazices and a Roman praefectus in chains to Caesarea: so they must have been operating in the Mazices' territory while attempting to capture Vincentius; it is therefore probable that Vincentius had there gone over to Firmus' side when he defected. This may have taken place in the country around Castellum Tingitanum (= Orléansville), cf. 29.5.25.

47. For examples of Theodosius' ferocious discipline see 29.5.20-23, 24, 31, 49, and Thompson, Historical Work, 90-91.

48. It should be pointed out here that Firmus' supporters were not only the tribesmen (as 29.5.28) but also included potentes municipes who must have been landowners in the frontier areas (29.5.43; see Thompson, Historical Work, 129f), and a number of Roman army units, the equites quartae sagittariorum cohortis and Constantianorum peditum pars (5.20); the tribunus of the former was executed for his tacit support of the rebel (5.24); Fericius, praefectus of the Mazices (probably gentiles assisting in the defence of the limes: Jones, LRE, [Later Roman Empire] 652), had actually lent support (5.21). Ammianus fails to make further mention of the municipes and their role in the revolt; perhaps as a curialis himself he felt impelled to defend them by silence.

49. Theodosius' handling of the locally based troops was exceptionally severe—its ruthlessness is unparalleled anywhere else in Ammianus: cf. the references given above. The discipline and effectiveness of these troops seems to have been poor for a number of years before Theodosius' arrival, cf. the lack of action by the corrupt Romanus (28.6 passim); the corruption surrounding the administration of supplies to the limes (Warmington, ‘Career of Romanus’, 56; Saumagne, ‘Un tarif fiscal au IV^e siècle de notre èré, in Karthago 1, 1950, 153ff); the despatch of a military mission by Valentinian in 365, probably to give a report on the military effectiveness of the North African provinces (26.5.14, and see Warmington, art. cit., 58); finally bribery and other financial skulduggery with military personnel of and by the tribunus et notarius Palladius sent out to pay stipendium debitum to the soldiers based in Africa (28.6.12). On the nature of this stipendium, see Jones, LRE, 624 and n32. It is possible that although it may have been a donatiuum it may equally well have been backpay, neglected for a period through the general corruption. If it is back-pay, it would be consonant with the general state of army discipline in Africa at this time (cf. 27.9.1-3) and with the demands made by Romanus on the Tripolitanians. Demandt, ‘Die afrikanischen Unruhen unter Valentinian I’, Afrika und Rom in der Antike, 277-292, is particularly useful for his analysis of background conditions here.

50. E.g. the fundi Petrensis, owned by a brother of Firmus and built in modum urbis (29.5.13, and shortly before Theodosius' move inland), and Gaionatis, muro circumdatum ualido (29.5.25). Fortified farms formed part of the frontier system in North Africa, being found on both sides of the fossatum; see Jones, LRE, 652 and n. 103, where he makes reference to Baradez, Fossatum Africae: recherches aériennes sur l'organisation des confins sahariens à l'époque romaine, where air photographs of selected areas show these farms and their relation to the limes in great detail. See also Goodchild and Ward-Perkins, ‘The Limes Tripolitanus in the light of recent discoveries’, JRS [Journal of Roman Studies] 39, 1949, 81-95.

51. At 30.7.10, Ammianus mentions, as contributory causes, military abuses; but he makes less of this aspect. See Warmington, art. cit., 61.

52. These are named at 29.5.11, 17, 27, 37, 40, 44, 50 (= a total of 10).

53. Stevens, ‘Hadrian and Hadrian's Wall’, Latomus 14, 1955, 394f.

54. TLL [Thesaurus Linguas Latinas] s.v.

55. Richmond, Roman and Native in North Britain, 114ff.

Select Bibliography

Baradez, J., Fossatum Africae: recherches aériennes sur l'organisation des confins sahariens à l'époque romaine, Paris 1949.

Demandt, A., ‘Die afrikanischen Unruhen unter Valentinian I’, Afrika und Rom in der Antike, Halle 1968, 277-292.

Dillemann, L., La Haute Mésopotamie orientale et les pays adjacents, Inst. Fr. d'Arch. de Beyrouth, 1962. (HMPA)

Dillemann, L., ‘Ammien Marcellin et les pays de l'Euphrate et du Tigre’, Syria 38, 1961, 87-155. (‘AMPET’)

Ensslin, W., ‘Zum Heermeisteramt des spätrömischen Reiches’, Klio 5, 1929, 306-325; and Klio 6, 1930, 102-147.

Grosse, R., Römische Militärgeschichte, Berlin 1920.

Jones, A. H. M., The Later Roman Empire, Oxford 1964. (LRE)

Laur-Belart, R. ‘Late lines from Basil to the Lake of Constance’, Congress of Roman Frontier Studies, 1949, 63ff.

Müller, A., ‘Militaria aus Ammianus Marcellinus’, Philolongus64, 1905, 503ff.

Nischer, E., ‘Das römische Heer und seine Generale nach Ammianus Marcellinus’, Hermes 63, 1928, 430-456.

Norden, E., Altgermanien, Leipzig 1934.

Richmond, I., Roman and native in North Britain, Edinburgh 1958.

Rosen, K., Studien zur Darstellungstkunst und Glaubwürdigkeit des Amminaus Marcellinus, (Diss.) Heidelberg 1968.

Saumagne, C., ‘Un tarif fiscal au IV^e siècle de notre ère’, Karthago 1, 1950, 108-200.

Seeck, O., Regesten der Kaiser und Päpste, Stuttgart 1919.

Stevens, C. E., ‘Hadrian and Hadrian's Wall’, Latomus 14, 1955, 384ff.

Thompson, E. A., ‘Constantine, Constantius II and the lower Danube frontier’, Hermes 84, 1956, 372-381.

Thompson, E. A., The Early Germans, Oxford 1968.

Thompson, E. A., The Historical Work of Ammianus Marcellinus, Cambridge 1947.

Thompson, E. A., The Visigoths in the time of Ulfila, Oxford 1966.